http://www.fao.org/globalsoilpartnership This presentation was made during the GSP/E-SOTER Global Soil Information workshop that took place at the FAO HQ, Rome , Italy from 20-23 March 2012. This presentation was made on the first day by Freddy Nachtergaele & Harrij van Velthuizen, and it presents The Harmonized World Soil Database (HWSD) A Global Soil Information System. ©FAO: http://www.fao.org

1. The Harmonized World Soil Database (HWSD)
A Global Soil Information System.
Freddy Nachtergaele
freddy_nachtergaele@hotmail.it
Harrij van Velthuizen
velt@iiasa.ac.at

2. INTRODUCTION
The Food and Agriculture Organization of the United Nations (FAO)
and the International Institute for Applied Systems Analysis (IIASA)
took the initiative of combining the regional and national updates of
soil information with the information already contained within the 1:5
M scale digital FAO-UNESCO Digital Soil Map of the World (DSWM)
into a new Harmonized World Soil Database (HWSD)
This approach was actively supported by the European Soils Bureau
Network (ESBN), ISRIC_World Soils and the Chinese Academy of
Sciences.
A first version of HWSD was published in 2008. The present version
(v1.2) contains geographical expansions and incorporation of new and
more detailed data and corrections for a number of soil parameters,
notably for Soil Bulk Density values.

3. APPROACH
1. HWSD uses as a basis soil legacy data and maps. It harmonizes this information
by using a standard grid and a single reference soil classification.
2. HWSD derives soil characteristics based on actually measured soil properties
for each soil type in the mapping unit, using statistical techniques (pedo/taxon
transfer functions).
3. HWSD makes optimum use of legacy soil data and has relied for collecting
these on national soil institutes and international organizations. FAO has a
track record of providing technical assistance to member countries to improve
their exsisting national soil datasets.
4. HWSD estimates a large number of soil properties (20) both for top- and
subsoil. In addition it gives soil structural and soil management information
(soil phases) and contains a number of associated datasets.
5. HWSD is an operational global soil database, although not all regions are yet
covered with the same reliability. All data, the manual and results are freely
available on the internet:
http://www.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML/
6. A new on-line HWSD viewer is being developed by ISRIC with other
partners.

4. ACHIEVEMENTS TO DATE
1. The HWSD already constitutes improvements for about 60% of
the land area as compared to the FAO/UNESCO Soil Map of the
World.
2. The Global Earth Observation System of Systems (GEOSS), has
identified an improved HWSD as an intermediate product to be
completed in the short term. (GEO 2009-2011 Work plan, 2009).
3. The HWSD is used to refine global land degradation findings
(GLADIS), and is an essential layer in the global land evaluation
approach for food security as undertaken by FAO/IIASA with
Global Agro-ecological zoning (GAEZ).
4. HWSD has already provided improved estimates of soil moisture
storage capacities and soil organic carbon pools in support of
climate change studies.
5. HWSD is likely to remain for some time to come the only
available global soil database. Further improvements both
geographically as quality wise are still required.

5. Application: Soil moisture holding capacity
(Source: Vincent van Engelen, ISRIC)

6. Application: Soil Suitability for rainfed agriculture ( at low input)

7. LEGACY SOIL DATA SOURCES
Legacy Soil Maps/Spatial databases
•(1) The European Soil Database (ESDB) extended with information of the
Northern Circumpolar soil map at 1:1 M scale. This database is considered of
moderate reliability with an adequate scale but often lacking soil profile
information.
•(2) The new Soil Map of China at scale 1:1 M produced by the Chinese
Academy of Sciences. The database is considered of moderate reliability for
the same reasons as the one above.
•(3) The SOTER databases mainly for Eastern, Central and Southern Africa,
South America and the Caribbean and parts of Asia. National SOTER
databases such as available for Senegal, Tunisia etc…This part is considered of
moderate to high reliability.
•(4) For the areas not covered by the above, mainly West Africa, North
America, South Asia and Australia, the FAO/UNESCO Digital Soil Map of the
World was re-interpreted. This part of the database is considered of relatively
low reliability.
Soil profiles
All profiles collected in the WISE database (ISRIC), the ones included in the
national and regional SOTER products, those of the European Soil Database
(JRC) and the National China soil profile dataset have been used (more than
12 000 soil profiles in total)

8. Main Legacy Soil Data sources in HWSD
ESDB
CHINA
SOTWIS
DSMW
No Data

9. Structure of the Relational Database
• As the product has as its main aim to be of practical use to environmental modelers
and serves perspective studies for agriculture through agro-ecological zoning, food
security planning and climate change impacts (among others), a raster map with
resolution of about 1 km (30 arc seconds by 30 arc seconds) was retained as the
basic unit allowing easy integration with other environmental databases.
• The database consists of 21600 rows and 43200 columns, of which 221 million grid
cells cover the globe’s land territory. Over 16000 different soil mapping units are
recognized, which are linked to harmonized attribute data which are of five kinds:
 General information on the soil mapping unit composition;
 Information related to soil phases;
 Physical and chemical characteristics of topsoil (0-30 cm) and subsoil (30-100 cm).
 Derived soil information
 Interpreted and related soil information
• Use of this standardized structure allows linkage of the attribute data with GIS to
display or query the composition in terms of soil units and the value of selected soil
parameters. A viewer is an integral part of the HWSD, making it also accessible to
non-GIS users. ISRIC and partners are developing a new on-line viewer.

10. 30 arc sec Grid
Morphological soil information Chemical soil properties Physical soil properties
Derived soil
properties
Soil depth Organic Carbon Sand fraction CEC clay fraction
Presence of Gelic properties pH(H2O) Silt fraction Base saturation
Presence of Vertic properties CEC soil Clay fraction ESP
Other Soil Phase information
Total Exchangeable
Bases Bulk Density USDA Texture Class
Soil Drainage class Calcium carbonate Reference Bulk Density
Gypsum
•ECe
5 min Grid
Derived Soil Qualities for Agriculture Soilscape Land Cover and Land Use Population
Nutrient availability, Slope %
Rain-fed Cultivated
land Rural population
Nutrient retention capacity, Slope Aspect
Irrigated Cultivated
land Urban Population
Rooting conditions Forest Land
Oxygen Availability Pasture Land
Excess Salts Barren/Slightly Veg. land
Toxicity Open Water
Workability Urban land
Content of the database: soil & related properties

11. Example of derived information included in HWSD
(Soil Nutrient availability)

12. From Soil Unit to soil property estimates:
Pedo- & Taxo-transfer functions
The mapping unit information in the Digital Soil Map of the World has
been linked to respectively topsoil and subsoil parameters derived from
the WISE soil profile database (Batjes et al., 1997 and Batjes, 2002).
This linkage was established through the soil unit classification (that is
why the term taxo-transfer is used) further subdivided by three topsoil
texture classes and the depth class.
For each combination occurring in the regional cluster MEDIAN values
for each property were determined and associated with the Soil Unit in
the mapping unit.
The number of profiles in the regional cluster in the WISE database
determines the reliability of the estimate (wnen n>30 = high reliability)

13. Pedo- and Taxo-transfer functions (2)
In the SOTER-derived part of the database (SOTWIS) use is made of the typical soil
profiles contained in the mapped SOTER units in the countries complemented with
the soil properties from the WISE database for the same regional cluster where soil
property information is missing.
As in the for DSMW, MEDIAN values (after outliers have been eliminated) are used to
estimate missing soil properties within the mapping unit.
In the European Soil Database pedo-transfer functions were used that derived values
for soil properties not measured from measured ones and ancillary information (for
instance: parent material).
Soil property estimates are of a higher reliability in SOTER areas than in other
geographical areas because the proportion of measured profile information is
significantly higher. This is illustrated in the next slide where the proportion of the
synthetic profile information used in the country is mapped.

14. Proportion of synthetic profiles used in SOTWIS
Example Senegal

15. HWSD Development options (1)
The HWSD is the only currently available digital global soil database and will remain so for
some time to come, but improvement are required in the quality and quantity of the present
data, their geographical coverage and the harmonization process.
(1) Improved Geographical coverage
(a) A large part of the HWSD coverage that still uses the FAO digital soil map of the world
(40%) is located in the industrial world. A commitment of these countries (the largest being
Canada, USA, Mexico and Australia) to produce data in the required format would result in
significantly enhanced product. This could be undertaken at low cost for instance in
universities using the existing excellent national soil databases.
(b) Updates in West Africa; Asia, North Africa and the Middle-East are required (largest
countries: India, Iran, Nigeria, Saudi Arabia). This would require modest investments and
resources channeled to national institutes to produce the required data in the required
format.
(c) Improvements in soil units as well as in soil properties in all areas would benefit from new
SOTER products such as eSOTER and other SOTER products planned in the future.

16. HWSD Development options (2)
(2) Improved Quality of Soil Property predictions
The overall accuracy of taxo-transfer and pedo-transfer functions can be improved
by (1) increasing the number of soil profiles used in the prediction and by (2)
correcting the soil property predictions for local conditions as done in SOTER.
In this respect the ongoing efforts of Globsoil.net project in the systematic
collection of soil profile data and the ongoing SOTER studies, which include soil
profile data are of particular interest. A mechanism to exchange information
between the three approaches should be established.
Nevertheless, one should warn against high expectations of users in the achievable
accuracy of soil data: Soils are dynamic systems and their properties vary
considerably in space and time. It is therefore unrealistic to expect an extremely
high accuracy, whatever approach is taken. Also inter and across laboratory
analytical results show a high variability

17. HWSD Development options (3)
(3) Improved Harmonization
Harmonization of the units and boundaries can be improved mainly by using the latest soil
classification scheme (at the moment the World Reference Base) as a standardized approach.
This is not a simple exercise as it would require a systematic re-classification of soil profiles
and soil types in WRB and the consequent statistical re-analysis of soil properties in order to
develop new taxo-transfer functions.
Harmonization of soil laboratory methods and quality control is another issue that may need
to be tackled in the long run in order to improve harmonization.
These harmonization issues will benefit from the findings and recommentdations of the
Global Soil Partnership.

18. Thank you